As modern power supplies edge upward in operational frequency, engineers have started to migrate to high-frequency power switch and rectifier technologies. The traditional planar or trench MOSFET switches with rise/fall times 30nsec to 60nsec are giving way to power switches such as superjunction MOSFETs, GaN MOSFETs, SiC MOSFETs and SiC Schottky rectifiers that switch in less than 5nsec.

To view such fast transitions, you typically need an oscilloscope with at least 1GHz bandwidth. Unfortunately, most commercially available voltage and current probes are woefully inadequate at these high frequencies. The average oscilloscope probe has a bandwidth of less than 300MHz. Current probes can have bandwidths of 60MHz to 100MHz or less. Furthermore, high-frequency voltage probes often cost over $12,000 and slightly better current probes start at $4,000. For power engineers who work for mid-sized companies, there is only one path: build your own probes. Designing and building high-frequency voltage and current probes requires a good understanding of RF, parasitics, transmission-line theory and field theory.

Commercial probe shortcomings

Commercially available oscilloscope voltage and current probes are robust, ergonomically well designed and accurate. They have served their markets well where the overwhelming number of applications operates at much less than 1GHz. The operating frequencies and the edges of new-generation switching transistors are exceeding 1GHz, resulting in rise and fall times in the sub 5 ns range.

A commercial probe's low bandwidth can create a major limitation to accurate measurements. Engineers often take for granted the slow rise and fall times and they can easily overlook missing information. In addition, the common probe's connection to the signal source can cause distortions. These connections have a significant length of unshielded connecting leads, particularly the ground lead. A 4–6in (10–15cm) ground lead can pick-up radiated noise from the circuit or other sources and inject it into the coax cable as a common-mode signal. This unrecognised noise adds to the real signal.

Figure 1 shows a typical commercial voltage probe. It contains a length of unshielded signal or ground wire that acts as a loop antenna. The amount of noise it picks up is proportional to the loop size and the amount of noise energy and noise spectrum. You can view this noise by simply clipping the ground lead to the probe tip and hold it near the target circuit board.

20170810_EDNA_D3Semi_Probes_01 (cr) Figure 1: Common voltage oscilloscope probe construction has a ground lead that you clip to the circuit under test.

Instead, you can construct your own 50Ω voltage probe. By constructing custom 50Ω voltage probes, you can better define and understand what is really happening within the circuit. The overall goals of constructing 50Ω voltage probes are:

  • Construct a known, quiet, high frequency signal path from the circuit to the oscilloscope.
  • Provide shielding as much as practically possible along the signal's path.
  • Gain the ability to control as many parasitic influences as possible.

1:1 Shielded Coax Voltage Probe

For those signals below the maximum input voltage rating of the oscilloscope input, you can use a cut length of a 50Ω BNC coax cable as your probe. The length of the unshielded centre conductor and the shield pigtail should be kept to less than 1in (25cm) to minimise noise pickup. For viewing a signal at a particular node, solder the centre conductor directly to that node; the ground lead should be soldered to the closest associated ground. That is, not to a ground that has a long PCB trace length between the probe and the node of interest. This probe only provides high frequency signal shielding from the target circuit to the oscilloscope. The input termination setting of the oscilloscope scope should be 1MΩ. Figure 2 shows the design of a 1:1 shielded probe.

20170810_EDNA_D3Semi_Probes_02 (cr) Figure 2: The 1:1 shielded voltage probe is based on coax cable. Inductance on the probe tip (LUS) and ground lead (LG) will limit bandwidth but the small size will help minimise noise pickup.

n:1 50Ω voltage probe

The n:1 probe is intended for signal amplitudes (including any spikes) that exceed the maximum voltage rating of the oscilloscope's input amplifier. This probe is a bit more complicated to construct. Its simplified schematic is shown in Figure 3.

20170810_EDNA_D3Semi_Probes_03 (cr) Figure 3: Simplified schematic of the n:1 voltage probe shows a series resistor RS that requires some calculations to find its value.

This brings us to the first and important step of determining the value of the sense resistor (RS). This is not as straightforward as you might think. There are several factors which you should take into consideration.

Set the oscilloscope's input termination to 50Ω. The oscilloscope's internal 50Ω terminating resistor becomes the bottom resistor of a resistor-divider circuit. You can safely assume that this resistor has better than a 0.1% tolerance. Its power dissipation should not exceed 0.25W. That power rating sets the maximum current that can enter the oscilloscope's input.

Additional considerations include:

  • Maximum amplitude of the signal across the 50Ω terminating resistor.
  • Power dissipated within the series sense resistor (RS).
  • Loading on the input circuit.

All of these considerations must be balanced among each other and they will dictate the gain setting of the oscilloscope input amplifier. If the signal is too low, the oscilloscope input gain must be set in the <100mV range. The displayed signal becomes noisy because the input signal is very close to the input amplifier's noise floor. This noise results in a reduction in ADC input resolution. The signal may only be acquired by the ADC's lowest four bits of the ADC (assuming an 8-bit ADC). You'll end up seeing the quantisation steps of the leas-significant bits (LSBs). This is somewhat unavoidable, especially in probes with a high step-down ratio. Figure 4 shows typical display of a 1000:1 50Ω probe.

20170810_EDNA_D3Semi_Probes_04 (cr) Figure 4: Low-level oscilloscope traces often show quantisation noise on input signals.

Figure 5 shows the basic construction of an n:1 voltage probe.

20170810_EDNA_D3Semi_Probes_05 (cr) Figure 5: The basic construction of an n:1 50 Ω Probe includes a 1/4 W resistor near the tip.

Follow these steps when designing the n:1 probe.

First, determine the resistor reduction ratio desired to result in an oscilloscope signal amplitude (including spikes), for the desired channel gain setting. It is typically nice to choose a decade-multiple resistor reduction ratio, since the displayed v/div setting differ only in the placement of a decimal point from the input voltage.

20170810_EDNA_D3Semi_Probes_eq01 (cr)

The typical input amplitude should not exceed the power rating of the internal input 50Ω terminating resistor. To produce the desired channel voltage, a current must pass through the 50Ω terminating resistor.

20170810_EDNA_D3Semi_Probes_eq02 (cr)

The power must be less than the power rating of the terminating resistor:

20170810_EDNA_D3Semi_Probes_eq03 (cr)

Calculate the value of the sense resistor (R1) by:

20170810_EDNA_D3Semi_Probes_eq04 (cr)

Now check the power dissipation of the sense resistor.

20170810_EDNA_D3Semi_Probes_eq05 (cr)

Check for the loading of the circuit you want to view. Here, you must understand and determine the effects upon the targeted circuit. If the probe draws too much sense current, then the probe will change (sometimes drastically) the operation of the target circuit. A general rule of thumb is:

20170810_EDNA_D3Semi_Probes_eq06 (cr)

There are instances where the initial considerations are met, but the probe overloads the target circuit. In that case, you must go back to step 1 and use a sense current lower than the current originally selected.

 
Next: Noise considerations for DIY oscilloscope probes »